Does anyone track factors for Bernoulli numbers?

[VBCurtis]

I decided that trying to find an irregular prime would be a top-5000 worthy way to learn to use Primo. It seems that to find one, one ECMs Bernoulli numerators in hopes for a prp cofactor.

Is there any documentation of how much ECM has been done on Bernoulli numbers above index 300? Does anyone keep the factors found? I found bernoulli.org, which has up to index 10000 but only trivial factors (under 1M); and Sam Wagstaff's page, which only tracks up to index 300.

In case I have success with the ECM portion, how does Primo scale with size of input? Is a 10,000 digit input (or even 5000) too big for a 6-core Haswell?

[paulunderwood]

Originally Posted by VBCurtis

In case I have success with the ECM portion, how does Primo scale with size of input? Is a 10,000 digit input (or even 5000) too big for a 6-core Haswell?

ECPP is O(log(n)^4). I guess a 10k digit would take about 10 days on a Haswell 6 core. Maybe less. Suck it and see.